In reality, the anxiety over Williamson's hiring is revealing a fissure in how we think opinion-column space should be distributed. We might think of two different rationales for why a magazine should publish a particular opinion columnist or column -- a reflective rationale and a legitimation rationale.
Under the reflective view, opinion columns should roughly mirror -- reflect -- the opinions that exist in the wider public around politics. We should have conservative columnists because many people are conservative, and it's important that the readership be exposed to that common view. Note that this makes no claim about whether this perspective is a good one, only that it's a common one.
Under the legitimation view, by contrast, opinion columnists should be chosen because the editors conclude that their views are affirmatively worth considering. This almost certainly falls short of requiring agreeing with the views -- I can think of many opinions I don't share but which I nonetheless think are worth wrestling with -- but it does represent some commentary on quality. Often (not always), the legitimation rationale for hiring a columnist cuts in the opposite direction from the reflective view -- as in the claim that so-and-so's perspective is one "you don't hear often" or is "novel and original" (though these are not in themselves reasons for publishing someone unless they're also paired with the ideas actually being decent. A string of gibberish isn't heard often in august editorial sections, but that doesn't mean it should be published).
One problem that's partially being illuminated by the Williamson hiring is that we're not sure what purpose we want opinion section editors to adopt. Some of the critics of Williamson's hiring note that, as a Never-Trump Republican, he's actually not reflective of conservative ideology as it's currently practiced -- wouldn't ideological diversity better be served by hiring an avowed Trump defender? But of course one can feel the anxiety in that sentence even as it's written -- do we really want the views of avid Trump supporters (Muslim ban, "shithole countries", and all) gracing more prominent media outlets? No, because we think it would legitimate them.
But if Williamson can't be defended as a reflective hire -- and a #NeverTrump Republican by definition isn't reflective (if even a fifth of American conservatives were of this ilk, Hillary Clinton wins in a landslide) -- he has to be justified based on legitimation. Yet it's really hard to justify him along that metric either. Certainly, the usual bromides about hearing "alternative points of view" won't do on their own. We need to know why "hang one quarter of American women" is the sort of keen social insight that is worthwhile on it's own right -- not simply as a "different" view but as an independently justifiable one.
So what do we want? The reflective case has one thing going for it, and that's that it would represent an honest portrayal of the state of conservative argumentation in America today -- which doesn't even carry Williamson's patina of #NeverTrump-ism. Yet going that route would have to mean consciously abandoning any legitimation justification for their publication -- and most editors don't really want to write that out of their job description. The fetishization of an objectively tiny branch of contemporary conservatism -- acting as if it is of more than trivial social influence -- is I think primarily grounded on how they let us deny what is before our eyes.
Seth Mandel suggests that one reason conservatives are skeptical of this reaction to Williamson is that -- even though some of the criticisms of him hold water in isolation -- it seems to happen to every conservative who joins a mainstream media outlet (cf. Bret Stephens). But the problem is that, given the state of conservatism today, an attempt to find a remotely representative yet not-terrible conservative voice will start to look like an error theory: it's possible in concept to imagine a representative conservative argument worthy of legitimation, but in practice none of them will qualify. Given where conservatism's center of gravity is right now, even if you step a standard deviation off the middle of the bell-curve, you're still likely to encounter at goodly chunk of the (again, wholly mainstream) conservative pathologies -- like "execute women" (Williamson) or "science is a hoax" (Stephens). But there's no reason -- at least once we stop grading conservatives on a curve -- why "I don't like Trump, but global warming is made up" should be thought to count as a good argument.